How to accurately predict the properties of molecules is an essential problem in AI-driven drug discovery, which generally requires a large amount of annotation for training deep learning models. Annotating molecules, however, is quite costly because it requires lab experiments conducted by experts. To reduce annotation cost, deep Active Learning (AL) methods are developed to select only the most representative and informative data for annotating. However, existing best deep AL methods are mostly developed for a single type of learning task (e.g., single-label classification), and hence may not perform well in molecular property prediction that involves various task types. In this paper, we propose a Task-type-generic active learning framework (termed Tyger) that is able to handle different types of learning tasks in a unified manner. The key is to learn a chemically-meaningful embedding space and perform active selection fully based on the embeddings, instead of relying on task-type-specific heuristics (e.g., class-wise prediction probability) as done in existing works. Specifically, for learning the embedding space, we instantiate a querying module that learns to translate molecule graphs into corresponding SMILES strings. Furthermore, to ensure that samples selected from the space are both representative and informative, we propose to shape the embedding space by two learning objectives, one based on domain knowledge and the other leveraging feedback from the task learner (i.e., model that performs the learning task at hand). We conduct extensive experiments on benchmark datasets of different task types. Experimental results show that Tyger consistently achieves high AL performance on molecular property prediction, outperforming baselines by a large margin. We also perform ablative experiments to verify the effectiveness of each component in Tyger.
Materials Property Prediction with Uncertainty Quantification: A Benchmark Study
Khoury College of Computer Science, Northeastern University(东北大学 Khoury 计算机学院);Department of Computer Science and Engineering, University of South Carolina(南卡罗来纳大学计算机科学与工程系)
On Graph Neural Network Ensembles for Large-Scale Molecular Property Prediction
University of Cambridge(剑桥大学);ML Collective(ML集体);Universitat Rovira i Virgili(罗维拉-威尔吉利大学);McGill University(麦吉尔大学);Wrocław University of Science and Technology(弗罗茨瓦夫理工大学);Obafemi Awolowo University(奥巴费米·阿沃洛沃大学)
Efficient Chemical Space Exploration Using Active Learning Based on Marginalized Graph Kernel: an Application for Predicting the Thermodynamic Properties of Alkanes with Molecular Simulation
Shanghai Jiao Tong University(上海交通大学);Lawrence Berkeley National Laboratory(劳伦斯伯克利国家实验室);Purdue University(普渡大学)
Robust Active Learning: Sample-Efficient Training of Robust Deep Learning Models
University of Luxembourg(卢森堡大学)
主动学习是一种已建立的技术,用于降低标记成本以构建高质量的机器学习模型。主动学习的一个核心组成部分是获取功能,它确定应该选择哪些数据进行注释。最先进的采集功能--以及更大程度上的主动学习技术--被设计成最大化干净性能(例如准确性),而忽略了鲁棒性,鲁棒性是一个越来越受到关注的重要质量属性。因此,主动学习产生的模型是准确的,但并不稳健。在本文中,我们提出了\emph{robust active learning},这是一个集成了对抗训练的主动学习过程--这是产生鲁棒模型的最成熟的方法。通过对11个采集函数,4个数据集,6个DNN架构和15105个训练DNN的实证研究,我们表明鲁棒主动学习可以产生鲁棒性(对抗性示例的准确性)范围从2.35%到63.85%的模型,而标准主动学习系统地实现了可忽略的鲁棒性(小于0.20%)。然而,我们的研究还表明,在准确性方面表现良好的采集函数在鲁棒性方面比随机采样差。因此,我们研究这背后的原因,并设计了一个新的采集功能,目标是干净的性能和鲁棒性。我们的采集函数-命名为基于密度的鲁棒采样熵(DRE)-优于其他采集函数(包括随机)的鲁棒性高达24.40%(3.84%,特别是比随机),同时保持竞争力的准确性。此外,我们证明了DRE是适用于作为模型再训练的测试选择度量,并从所有比较的功能中脱颖而出,高达8.21%的鲁棒性。
Generative Active Learning for the Search of Small-molecule Protein Binders
Mila – Québec AI Institute(米拉-魁北克人工智能研究所);Dreamfold;McGill University(麦吉尔大学);Université de Montréal(蒙特利尔大学);IRIC, Université de Montréal(蒙特利尔大学IRIC研究所);Valence Labs(Valence实验室);X-Chem(X-Chem公司);MIT(麻省理工学院);AI Redefined